Experimental Modeling For Actuator Pneumatic Position System

As one of three industrial automation based equipment (measurement instrument, the distributed control system (DCS) and actuator), the technical level of the actuator has a great influence on process industry production quality,efficiency,safety,energy saving and environmental protection and so . As a terminal execute equipment for process control system, system modeling and identification research on positioning system of the pneumatic actuator is a key theory for research on intelligent valve locator accurate posiontioning control algorithm,optimization design of the control valve,actuator fault diagnosis and proper preventive maintenance techniques as well as improvement on dynamic performance of process control system. In this paper, taking the pneumatic actuator as object, building the corresponding experimental platform, based on the measured data, first using least-square method for fitting curve, acquiring high-order system model with pure time-delay from system identification, then studying the existing problem of conventional genetic algorithm, and finally introducing the genetic algorithm into system model identification, respectively obtaining the pneumatic actuator intake and exhaust high-precison dynamic model.First introduce the basic principle of least-square method, deduce low order,high order continuous model parameter estimation formula, detailly introduce choose mothod of the delay time,calculation of the function of multiple integrals,time-domain error used for identification model evaluation and the basic process of parameter estimation. Then design and procduce least-square programs, achieve the precision dynamic model for the pneumatic actuators stem positive and reverse displacement and make validation for the models. According to identification results, it is concluded that when the pneumatic actuators intake and exhaust, stem positive and reverse displacement model are two order and three order delay model respectively. Finally study two experimental control parameters: sampling interval timeΔt and deadline t N for parameter estimation of influence, find in the step-response experiment if timeΔt is smaller, the incentive data is more, estimating values of the model parameters is more accurate.Convergence rate of conventional genetic algorithm is slow, and the evolutionary process for closed competition is easy to generate premature convergence. In view of the above question, mainly study fundamental principle of crossover operator and mutation operator and propose the improvement methods respectively. Improved crossover operator expands chromosome gene activity interval, maintains the diversity of genes, has random search capabilities, extensively search the solution space and can escape from local optimal solution. Mutation probability of improved mutation is not constant, similar to the variation step, increases with the numbers of evolution generation. In order to avoid damaging father generation excellent chromosomes due to high mutation probability, this paper also introduces back to insert operator based on fitness, whose basic idea: selection,crossover and mutation operators get all the offspring which are instead of inadaptable ones in the father generation individual to compose new populations. In the form of digital simulation, the same simulation object is carried out by domain response experiment and impulse response experiment and find the two identification models are the same as the simulation object. Finally by experimental method proof that even without enough incentive data, based on genetic algorithm identification method also can obtain the pneumatic actuator intake and exhaust high-precision dynamic model.